AYWUN A1-5000 - 5VSB nearly fixed

[socketa]

This PSU is the same as the one in the switchmode power supply article by c_heddge
I got it with a non-operational 5VSB
Quite a few damaged components - circled in the schematic (ignore the secondary side, i used a schematic from a different PSU, and only adjusted the 5VSB circuit)
The electrolytic cap was high ESR; so i think what happened is that it couldn't charge up enough to turn on the TO92 transistor, thus over-heating the main MOSFET and causing it to short, which, in turn, damaged the other components
The source of the MOSFET is connected to the base of the TO92 transistor, so it looks like it was also normally supplying some voltage to the TO92 base, but not enough to turn it on - and i think that the optocoupler circuit provided the bit of extra voltage to turn it on.

The photo in the article was the only one that i could find to give me an indication of the value of the burned of resistor that is next the the middle (driver?) transformer (pretty sure that it's 620 ohms)

After replacing what i thought were all of the faulty components the 5VSB came up but it was way too high, like 10V or more, IIRC
Then i found an open resistor between MOSFET source and the TO92 base, so am waiting for that to arrive.
AS you can see from the first resistor that i replaced, a 1/4W is a bit too long to replace a 1/8W; so i ordered 1/6W resistors

Also, you can see that the position of the first 3.3V output cap in not ideal - housed-in by the three toroids, like a mini oven? (i'd image that they would get fairy warm).
I'm going to test it, once i get the 5VSB oscillating to 5V

[socketa]

After replacing that 36ohm resistor, the 5VSB was 22V
So thought that there could be faulty components on the secondary
Replaced the opto with another one that caused an LED to glow brighter, and then checked the output caps, which were both good
Eventually found that the TL431 had low resistance (50Kohms-70Kohms)between reference and anode
I guess that the 5VSB got so high that it exceed this this regulator's specs
So after replacing it, the 5VSB is now 5V

Will now have a look at that 3.3V output cap, and then see if the unit will start up

[socketa]

That 3.3V output cap tests good, and it's rated for 16V, which i guess increases it's tolerance for heat.
There is some protection kicking in when i ground PS_ON wire - the rail voltages come up, but the PSU pretty much shuts down straight away.
The output rails aren't shorted, so i guess that the problem is more likely to be on the primary
Will start with checking the primary side...

[stj]

if the standby went to 22v, you should check the datasheets for the main chips - they are powered by the standby and may be damaged.
some years ago there was a series of threads about converting the standby side of some psu's to use a proper controller chip to prevent this type of crap.
i dont know if they survived the recent atachment loss thing.

[socketa]

Good point.
It's an equivalent TL494/LM339 setup.
The chips are probably ok, since the LM339 comparator is sending signal to the PWM controller dead time control pin, and shutting down the PSU.
At first i thought that the rail voltages were all within spec using a multimeter, but i can now see on the analog scope that the 12V is initially hitting 15V, 5V is hitting 6V, and 3.3V is hitting 4V.
The main switching transistors aren't shorted, so i wonder what's causing the overvoltage on all of the rails - I guess that the PWM part of the chip could now be faulty like you said - like maybe it's turning on, but not switching

[stj]

did you replace all the small electrolytics yet?

[socketa]

Not yet
I'll test the other small elecrolytics on the next fine day, and update with results
Will start with those two that look like they provide negative bias to the main switching transistors to turn them off faster

[socketa]

Replaced the PWM chip with a KA7500, which seems to be equivalent except for a reference voltage of 5.0V vs 4.9V for the original chip.
PSU still shut down.
So had a look around the LM339, and it all seemed to be good
Then thought that maybe the PWM is not getting correct voltage input from the 12V or 5V rails
So traced back to them (resistor in between each rail and the chip pin) and noticed high resistance of 40 ohms coming from this 5V rail trace.
The trace is damaged, and looks like another trace is also damaged.

[stj]

you need to re-enforce that or the crack will spread.
i usually deal with that by putting a piece of old fiberglass pcb on the top with epoxy.

[socketa]

Thanks for pointing that out - thought that it was just a scratch. So yeah, right you are - i looked at the top of the PCB, and the damage is visible on the top of the PCB, on both sides of the screw hole - maybe it was dropped at some stage.
I'll get around to doing that, and post an update here.

[socketa]

Reinforced corner (could have used a longer piece of PCB)
The PCB had a fair amount bend along it's length, so i managed to straighten it but applying some force and a heat gun (only took about 10s of heating). I used a zip tie to help, and then after realizing that the board responded well to the heat, and wanted to return to it's original flatness, i then just applied a bit of pressure at the ends while heating, to straighten them.
Pretty straight now.

Since this PSU would probably have being installed with the components hanging downward (Cases have the smaller screw hole separation distance on the top of the mounting space), maybe that is what caused the PCB to bend and then start cracking at the corner.
I checked a couple of other PSUs, and and they are also designed with the same mounting orientation, which seems wrong to me, as it causes the board to bend because of gravity, and the plastic standoff that's between the primary and secondary would otherwise also act to prevent board-bending if the PSU mounting holes were the other way around - plus all of the heat rises upwards to the PCB, which, again, encourages board-bending.

Bridged the cracks with solder, after scraping the solder mask off of the two traces.
Then reapplied new solder mask.
Works nearly good, except the 3.3V rail quickly goes to 3.45V (5V rail goes from 5.12V to 5.16V, and 12V rail is good), with just a hard drive connected
I then put a 4.7ohm resistor across the 3.3V rail, and it started at 3.36V, and within a second (takes longer than before) it rose to 3.45V (5V rail, no change from aforementioned)

[socketa]

Also notice slight noise that's coming from around standby transformer when the PSU is powered up, but it pretty much goes away as the hard drive spins up

[socketa]

I just used another schematic in this post, because it's similar circuitry as this Aywun 3.3 magamp. (Had to add in next post)
My PSU has TL431 reference pin voltage divider of 680 and 1800 which would give 2.39V at ref pin
internal ref for TL439 is 2.495V
... so is is possible that this PSU was manufactured with the wrong resistor 680, instead of 580 - would it be easy for the person making this PSU to use blue grey brown instead of green grey brown?
Blue could look like green in poor lighting
This higher value resistor would cause the TL431 to turn on more and regulate at a higher voltage, because the PNP transistor would cause the magamp coil to saturate for longer?

[socketa]

[socketa]

I see that standard resistor values would be 560, or 620; so i replaced the 680 ohm with a 620 (actually 615) ohm, so that now the 3.3V rail voltage has decreased from 3.45V to 3.397V.
Don't know why they didn't use a 620 ohm resistor rather than a 680 ohm, which brings 3.3V rail closer to 3.30V
Anyhow, it's fixed now

[Agent24]

Aywun isn't exactly a great brand and 3.45v is still (just) within the ATX spec. So they probably just didn't care.

[socketa]

Aywun isn't exactly a great brand and 3.45v is still (just) within the ATX spec. So they probably just didn't care.

Yes, ...because when i use a standard 560 ohm resistor, the 3.3V rail becomes 3.32V
(5V rail is 5.177V)
Will leave it at that.

[Agent24]

They probably ran out of 560 Ohm resistors, and being cheap AF, just used the next best ones they had.